Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!gatech!news.sprintlink.net!EU.net!news2.EUnet.fr!news.fnet.fr!ilog!lepape
From: lepape@ilog.fr (Claude Le Pape)
Subject: Re: Problem: Constraint Based Scheduling
Message-ID: <1995May15.134526.1182@ilog.fr>
Reply-To: lepape@ilog.fr (Claude Le Pape)
Organization: ILOG S.A., Gentilly, France
Date: Mon, 15 May 95 13:45:26 GMT
Lines: 140


Subject: Re: Problem: Constraint Based Scheduling

>  I am working on my MSc thesis where I am examining some heuristics
> for Constraint based intelligent scheduling.  My implementation
> language is CLP(FD) and my chosen scheduling domain is the classic 
> "job shop" model.
>
>  I'm having some problems trying to scientifically "interpret" my 
> results.
>
>  For example, given that scheduling is NP hard, and assuming
> I'm using examples "large" enough that a proof of optimality
> is not feasable, how can I verify the effectiveness of a 
> heuristic? 

One thing you can do at least is try your heuristics on small-to-medium
instances. A set of 66 such instances, with up to 300 operations, is
available from the Imperial College OR library. Information on this
library is included below.

As of today, the optimal value of 62 out of the 66 instances is known.
However, no single program provides the optimal solution to all of the 62   
instances, so it is still a reasonably good benchmark to test heuristics.

13 of the 66 instances have been selected by Vaessens et al (reference
below) to test and compare local search algorithms. Currently, the
best is a tailored Tabu search method due to Nowicki and Smutnicki
with a mean relative distance to the optimal solution (or best proven
lower bound) of 0.35%. We recently obtained a mean relative distance
to the optimal solution of 0.72%, using ILOG SOLVER and ILOG SCHEDULE,
two complementary industrial constraint programming tools. Another
tabu search method, this one due to Dell'Amico and Trubian achieves
a mean relative distance of 0.82%. A LOT of heuristic methods fall in
the 2 to 5% range, so you can at least compare to those.

  Claude LE PAPE			net: lepape@ilog.fr
  ILOG S.A.				tel: +33 1 46 63 66 66
  2 Avenue Gallieni - BP 85		fax: +33 1 46 63 15 82   
  F-94253 Gentilly Cedex 		url: http://www.ilog.fr
  FRANCE				     http://www.ilog.com


----------
REFERENCES
----------

R. J. M. Vaessens, E. H. L. Aaarts and J. K. Lenstra.
Job-Shop Scheduling by Local Search.
COSOR Memorandum 94-05, Eindhoven University of Technology, 1994.

E. Nowicki and C. Smutnicki.
A Fast Taboo Search Algorithm for the Job-Shop Problem.
Preprinty nr. 8/93, Instytut Cybernetyki Technicznej, Politechnicki
Wroclawskiej, 1993.

M. Dell'Amico and M. Trubian.
Applying tabu search to the job-shop scheduling problem.
Annals of Operational Research, 41, 231-252, 1993.

----------
OR LIBRARY
----------

OR-Library is a collection of test data sets for a variety of
problem areas. These test data sets can be accessed via emailing
to o.rlibrary@ic.ac.uk a message containing the name of the required
file, or via ftp/WWW using the addresses at the bottom of this 
message.

The following table gives the relationship between problem
area and the appropriate file:

Problem area                                  File
Assignment problem                            assigninfo
Bin packing                                   binpackinfo 
Crew scheduling                               cspinfo
Data envelopment analysis                     deainfo
Frequency assignment problem                  freqassinfo
Generalised assignment problem                gapinfo
Graph colouring                               colourinfo 
Integer programming                           mipinfo
Linear programming                            lpinfo
Location:
     capacitated warehouse location           capinfo
     p-median - uncapacitated                 pmedinfo
     p-median - capacitated                   pmedcapinfo
     uncapacitated warehouse location         uncapinfo
Matching                                      matchinfo
Maximum clique                                maxcliqueinfo
Min-cut clustering                            mincutinfo
Multiple knapsack problem                     mknapinfo
Network flow                                  netflowinfo
Quadratic assignment problem                  qapinfo
Resource constrained shortest path            rcspinfo
Scheduling:
     flow shop                                flowshopinfo
     job shop                                 jobshopinfo
     open shop                                openshopinfo
Set covering                                  scpinfo
Set partitioning                              sppinfo
Steiner:
     Euclidean Steiner problem                esteininfo
     Rectilinear Steiner problem              rsteininfo
     Steiner problem in graphs                steininfo
Stochastic linear programming                 stochlpinfo
Three-dimensional packing                     thpackinfo 
Time series forecasting                       timeinfo
Travelling salesman problem                   tspinfo
Two-dimensional cutting:    
     assortment problem                       assortinfo
     constrained guillotine                   cgcutinfo
     constrained non-guillotine               ngcutinfo
     unconstrained guillotine                 gcutinfo
Vehicle routing:                   
     fixed areas                              areainfo
     fixed routes                             fixedinfo
     period routing                           periodinfo
     single period                            vrpinfo
     sparse feasibility graph                 vrpfeasinfo

Instructions on how to use OR-Library can be found in the file paper
or in the article J.E.Beasley, "OR-Library: distributing test
problems by electronic mail", Journal of the Operational 
Research Society 41(11) (1990) pp1069-1072.

J.E.Beasley, June 1990.
             Last update: February 1995.

All the files in OR-Library are available via anonymous ftp to
      mscmga.ms.ic.ac.uk
The numeric equivalent of this ftp address is 155.198.66.4

World Wide Web access is also available at http://mscmga.ms.ic.ac.uk/

-- 
  Claude LE PAPE			net: lepape@ilog.fr
  ILOG S.A.				tel: +33 1 46 63 66 66
  2 Avenue Gallieni - BP 85		fax: +33 1 46 63 15 82   
  F-94253 Gentilly Cedex 		url: http://www.ilog.fr
